Quaternary Vertebrate Faunas from Sumba, Indonesia: Implications for Wallacean Biogeography and Evolution Samuel Turvey University of Cambridge

University of Wollongong Research Online

Faculty of Science, Medicine and Health - Papers Faculty of Science, Medicine and Health

2017 Quaternary vertebrate faunas from Sumba, Indonesia: Implications for Wallacean biogeography and evolution Samuel Turvey University of Cambridge

Jennifer Crees University of Cambridge

James Hansford University of Southampton

Timothy Jeffree University of Cambridge

Nick Crumpton University of Cambridge

See next page for additional authors

Publication Details Turvey, S. T., Crees, J. J., Hansford, J., Jeffree, T. E., Crumpton, N., Kurniawan, I., Setiyabudi, E., Guillerme, T., Paranggarimu, U., Dosseto, A. & Van Den Bergh, G. D. (2017). Quaternary vertebrate faunas from Sumba, Indonesia: Implications for Wallacean biogeography and evolution. Proceedings of the Royal Society B: Biological Sciences, 284 (1861), 20171278-1-20171278-10.

Research Online is the open access institutional repository for the University of Wollongong. For further information contact the UOW Library: [email protected] Quaternary vertebrate faunas from Sumba, Indonesia: Implications for Wallacean biogeography and evolution

Abstract Historical patterns of diversity, biogeography and faunal turnover remain poorly understood for Wallacea, the biologically and geologically complex island region between the Asian and Australian continental shelves. A distinctive Quaternary vertebrate fauna containing the small-bodied hominin Homo floresiensis, pygmy Stegodon proboscideans, varanids and giant murids has been described from Flores, but Quaternary faunas are poorly known from most other Lesser Sunda Islands. We report the discovery of extensive new fossil vertebrate collections from Pleistocene and Holocene deposits on Sumba, a large Wallacean island situated less than 50 km south of Flores. A fossil assemblage recovered from a Pleistocene deposit at Lewapaku in the interior highlands of Sumba, which may be close to 1 million years old, contains a series of skeletal elements of a very small Stegodon referable to S. sumbaensis, a tooth attributable to Varanus komodoensis, and fragmentary remains of unidentified giant murids. Holocene cave deposits at Mahaniwa dated to approximately 2000-3500 BP yielded extensive material of two new genera of endemic large-bodied murids, as well as fossils of an extinct frugivorous varanid. This new baseline for reconstructing Wallacean faunal histories reveals that Sumba's Quaternary vertebrate fauna, although phylogenetically distinctive, was comparable in diversity and composition to the Quaternary fauna of Flores, suggesting that similar assemblages may have characterized Quaternary terrestrial ecosystems on many or all of the larger Lesser Sunda Islands.

Disciplines Medicine and Health Sciences | Social and Behavioral Sciences

Authors Samuel Turvey, Jennifer Crees, James Hansford, Timothy Jeffree, Nick Crumpton, Iwan Kurniawan, Erick Setiyabudi, Thomas Guillerme, Umbu Paranggarimu, Anthony Dosseto, and Gerrit D. van den Bergh

This journal article is available at Research Online: http://ro.uow.edu.au/smhpapers/4954 Downloaded from http://rspb.royalsocietypublishing.org/ on September 18, 2017

Quaternary vertebrate faunas from rspb.royalsocietypublishing.org Sumba, Indonesia: implications for Wallacean biogeography and evolution

Samuel T. Turvey1, Jennifer J. Crees1, James Hansford1,2, Timothy E. Jeffree1, Research Nick Crumpton1, Iwan Kurniawan3, Erick Setiyabudi3, Thomas Guillerme4, Cite this article: Turvey ST et al. 2017 Umbu Paranggarimu5, Anthony Dosseto6 and Gerrit D. van den Bergh7 Quaternary vertebrate faunas from Sumba, 1Institute of Zoology, Zoological Society of London, Regent’s Park, London NW1 4RY, UK Indonesia: implications for Wallacean 2Ocean and Earth Science, National Oceanography Centre, University of Southampton, Southampton SO14 3ZH, UK biogeography and evolution. Proc. R. Soc. B 3Geology Museum Bandung, Bandung 40122, Indonesia 284: 20171278. 4Department of Life Sciences, Imperial College London, Silwood Park, Ascot SL5 7PY, UK 5 http://dx.doi.org/10.1098/rspb.2017.1278 KOPPESDA, Kampung Arab, Waingapu, Sumba Timur, Indonesia 6Wollongong Isotope Geochronology Laboratory, School of Earth and Environmental Sciences, University of Wollongong, Wollongong, New South Wales 2522, Australia 7Centre for Archaeological Science, School of Earth and Environmental Sciences, University of Wollongong, Wollongong, New South Wales 2522, Australia Received: 7 June 2017 STT, 0000-0002-3717-4800; TG, 0000-0003-4325-1275 Accepted: 21 July 2017 Historical patterns of diversity, biogeography and faunal turnover remain poorly understood for Wallacea, the biologically and geologically complex island region between the Asian and Australian continental shelves. A distinc- Subject Category: tive Quaternary vertebrate fauna containing the small-bodied hominin Homo Palaeobiology floresiensis, pygmy Stegodon proboscideans, varanids and giant murids has been described from Flores, but Quaternary faunas are poorly known from most other Lesser Sunda Islands. We report the discovery of extensive new Subject Areas: fossil vertebrate collections from Pleistocene and Holocene deposits on palaeontology Sumba, a large Wallacean island situated less than 50 km south of Flores. A fossil assemblage recovered from a Pleistocene deposit at Lewapaku in the Keywords: interior highlands of Sumba, which may be close to 1 million years old, contains biogeography, island evolution, murid, a series of skeletal elements of a very small Stegodon referable to S. sumbaensis, Quaternary extinctions, Stegodon, Wallacea a tooth attributable to Varanus komodoensis, and fragmentary remains of unidentified giant murids. Holocene cave deposits at Mahaniwa dated to approximately 2000–3500 BP yielded extensive material of two new genera of endemic large-bodied murids, as well as fossils of an extinct frugivorous vara- Authors for correspondence: nid. This new baseline for reconstructing Wallacean faunal histories reveals Samuel T. Turvey that Sumba’s Quaternary vertebrate fauna, although phylogenetically distinc- e-mail: [email protected] tive, was comparable in diversity and composition to the Quaternary fauna of Gerrit D. van den Bergh Flores, suggesting that similar assemblages may have characterized Quaternary terrestrial ecosystems on many or all of the larger Lesser Sunda Islands. e-mail: [email protected] 1. Introduction Islands represent important ‘natural laboratories’ for investigating evolutionary patterns, processes and dynamics [1,2], and unusual patterns of morphological evolution are observed in island environments, including gigantism in small- bodied lineages and dwarfism in large-bodied lineages [3–5]. However, insular taxa are particularly vulnerable to human pressures, and many island systems have lost a substantial proportion of their Quaternary vertebrate faunas following prehistoric human arrival [6,7]. Reconstructing historical diversity is therefore essential to overcome this ‘extinction filter’ and understand the evolutionary origin, morphological adaptation and biogeographic affinity of island faunas. Electronic supplementary material is available Such research is particularly important in southeast Asia, a geologically com- online at https://dx.doi.org/10.6084/m9. plex region with spatially complex biodiversity and endemism, which has figshare.c.3844780. & 2017 The Authors. Published by the Royal Society under the terms of the Creative Commons Attribution License http://creativecommons.org/licenses/by/4.0/, which permits unrestricted use, provided the original author and source are credited. Downloaded from http://rspb.royalsocietypublishing.org/ on September 18, 2017

2 119° E 120° E rspb.royalsocietypublishing.org

1 rc .Sc B Soc. R. Proc.

10° S 3 284

Bali 20171278 : Flores Lombok Sumbawa Timor Sumba N 50 km 50 km

Figure 1. Quaternary fossil localities on Sumba. 1, Lewapaku; 2, Watumbaka; 3, Mahaniwa. Inset, location of Sumba in Lesser Sundas, showing direction of Indo- nesian Throughflow past Flores and nearby islands towards Sumba. provided insights into biogeographic processes for over a cen- Flores has never been joined to other large islands in the tury [1]. In particular, faunal relationships across Wallacea, Lesser Sundas, but its extant fauna shows biogeographic affi- the islands that were never connected with the Asian or Austra- nities with nearby islands [8,9], suggesting that ecologically lian continental shelves during periods of low sea level, similar Quaternary vertebrate faunas may also have had a challenge easy interpretation [8–10]. Some vertebrate groups wider former distribution across Wallacea. Distinctive extinct (e.g. proboscideans) are known to have become completely Quaternary vertebrates (mainly endemic Stegodon species, extinct in Wallacea during the Late Quaternary, possibly rodents and/or other small-bodied vertebrate taxa) have also owing to prehistoric interactions with human colonists [4,5] or been described from Sulawesi, Timor, Halmahera and some alternatively because of climatic factors or postulated long- smaller Wallacean islands [4,10,13,29–34]. In particular, two term genetic decline [11]. Other vertebrate groups (e.g. murid Stegodon species, fossil varanids and a series of extinct giant rodents) still exhibit high levels of species richness and morpho- murids have been reported from the Quaternary of Timor, logical disparity across the region, but the extent to which their although most of these rodent taxa have not yet been formally Quaternary diversity and biogeography have been modified by described [11,13,29,30,34]. However, the Quaternary record for past human activity remains unclear [12–14]. much of Wallacea is still very poorly known, limiting our Interest in Quaternary vertebrate evolution in Wallacea understanding of the evolutionary history and faunal relation- has been raised by the discovery of a small-bodied endemic ships of the region’s insular biotas [4,10]. Research into the hominin, Homo floresiensis, from the Pleistocene of Flores in Quaternary record of other Wallacean islands would, therefore, the eastern Lesser Sunda Islands [15], with hominin bones help to characterize patterns of past diversity and turnover and stone artefacts dating from approximately 1.02 Ma until in insular vertebrate faunas in relation to different island approximately 100–50 ka [16,17]. H. floresiensis coexisted with environments, as well as assisting reconstruction of regional a distinctive, impoverished Quaternary insular vertebrate colonization histories, evolutionary processes and island fauna. Late Pleistocene deposits on Flores contain the ecologies [3,22,35]. pygmy proboscidean Stegodon florensis insularis, the Komodo Sumba (figure 1) is a large island in the Lesser Sundas, dragon Varanus komodoensis and a smaller extinct varanid with an area of approximately 11 000 km2, greater than 70% with distinctive bunodont teeth, Varanus hooijeri, giant murid that of Timor and greater than 80% that of Flores [8]. It is a rodents (Papagomys armandvillei, Papagomys theodorverhoeveni, continental fragment or ‘microcontinent’ in the non-volcanic Spelaeomys florensis and an undescribed giant shrew rat), Sunda–Banda forearc system, and has never been con- smaller murids (Komodomys rintjanus, Paulamys naso, Rattus nected to other Lesser Sunda Islands. It has occupied hainaldi) and a giant stork [12,18–24]. Older Pleistocene depos- approximately its present position since the Miocene, and its on Flores also contain a giant tortoise, another giant has been subaerial from the Late Pliocene [36–39]. Sumba con- murid (Hooijeromys nusatenggara) and other Stegodon species tains several extant endemic vertebrates, including eight [12,20,22]. All three smaller murids, one giant rat endemic birds [40,41], and shares other restricted-range ende- (P. armandvillei) and Komodo dragons still occur on Flores mics with Flores [8,40]. Mammal research conducted on and/or nearby small islands [12,24–27]; P. theodorverhoeveni, Sumba has been limited [42,43], and its Quaternary history is Sp. florensis and V. hooijeri survived into the Holocene but are poorly known. The only Quaternary vertebrate fossil reported now extinct [20,22,23,28]. from Sumba is a Stegodon mandible found in 1979 in a coastal Downloaded from http://rspb.royalsocietypublishing.org/ on September 18, 2017

terrace near Watumbaka and described as an endemic species, agent. The large quantity of rodent material and observed skeletal 3 S. sumbaensis [44]. representation in a concentrated surficial bone layer is consistent rspb.royalsocietypublishing.org We conducted new palaeontological research on Sumba in with prey accumulation by an owl or other raptor, probably October 2011 [45] and June–July 2014, and recovered extensive feeding in the entrance chamber [47]. Fossil material occurred at new vertebrate collections from Pleistocene and Holocene much lower density in Liang Minabuti, where accumulation is interpreted as non-predator mediated. deposits. We describe these finds here in detail, and use them to refine our understanding of Quaternary evolution and biogeography in Wallacea and to provide a new baseline for 3. Results reconstructing the region’s faunal history. Measurements were taken using dial callipers accurate to 0.01 mm. Rodent tooth measurements follow [23], and

2. Material and methods rodent molar crown nomenclature follows [13]. All previou- B Soc. R. Proc. sly undescribed fossil material reported here is accessioned (a) Pleistocene fauna in the Vertebrate Palaeontology Collection, Geology The Watambuka type locality of S. sumbaensis (figure 1), where Museum Bandung. the holotype mandible was found embedded in a hard calcified sandstone and gravel deposit cropping out at the surface of a dry (a) Stegodon sumbaensis Sartono 1979 284 river bed [44], was visited in 2011. This shallow dry river valley 20171278 : The holotype of S. sumbaensis consists of a partial left mandib- lies at the foot slope of a 3 m high cliff face exposing a 2–3 m ular body with two molars [44]. Re-examination of the thick transgressive sequence of beach rock sand with reworked coral rubble capped by coral-reef limestone. No additional specimen by G.D.v.d.B. shows that it is an adult specimen vertebrate fossil remains were recovered. with a total height (excluding broken condyle) of approxi- Vertebrate fossils were collected at Lewapaku village in 2011, mately 250 mm, a maximum width of the horizontal ramus from a 2.5 m deep, 2 2 m excavation in the backyard of a local of approximately 60 mm (measurement M18 [48]; estimated resident (9.6978 S, 119.8738 E) (figure 1; electronic supplementary because medial side of ramus has superficial damage of cortical material, figure S1). The site is located in a 8 12 km wide basin bone at the point of maximum thickness), and with the two at 500–520 m a.s.l. and surrounded by karst hills. The basin floor last molars (m2–m3) present and showing dental wear stage is made of horizontally bedded, well-consolidated marine sandy M2–M3-A of [49] (modified in [48]; electronic supplementary marl of the Late Miocene–Pliocene Waikabubak Formation [46], material, text S1). Despite some damage to the specimen, it is overlaid by a 20 cm thick sandy layer containing terrestrial ver- clearly significantly narrower than adult mandibles of Stegodon tebrate remains and then by 2.20 m of homogeneous lacustrine/ sondaari from Tangi Talo on Flores with similar wear stages paludrine clay (electronic supplementary material, figure S1). The fossil layer consists of poorly sorted coarse sandy material (e.g. MGB-TT12-TF-F80, wear stage M2–M3-A, maximum (carbonate fragments with local origin; pebbles up to 8 cm diam- width ¼ 74.5 mm; MGB-TT13-TG-F1, slightly younger but eter) resting unconformably on the marine marl. Larger fossils larger individual with wear stage M2-A (i.e. no m3 ridges were collected directly from the unconsolidated sediment, and showing wear), maximum width ¼ 81 mm [5]). Dental pro- the sediments of the fossil-bearing layer were wet-sieved over portions shown by S. sondaari are also greater than shown by 2 mm mesh. A series of skeletal elements of a small Stegodon the holotype of S. sumbaensis [5]. As S. sondaari was to date (apparently from multiple individuals based on ontogenetic vari- the smallest known well-described Stegodon, these measure- ation between elements), a large ziphodont varanid tooth, and a ments show that the S. sumbaensis type mandible clearly rodent lower incisor, two proximal femur fragments and tibial represents an extremely small pygmy Stegodon species. diaphysis fragment were collected. The fossils from Lewapaku that can be attributed to a small-sized Stegodon are a juvenile mandible (MGB-19650), a (b) Holocene fauna worn molar fragment (MGB-19651), an isolated molar lamella Vertebrate fossils were found in 2014 in two small caves in karst (MGB-19652), a portion of a tusk (MGB-19653), a proximal hills near Mahaniwa village at approximately 830 m a.s.l. metacarpal III fragment (MGB-19654), a distal metapodial (figure 1; electronic supplementary material, figure S1). Most fos- fragment (MGB-19655), a proximal humerus fragment sils were found in Liang Lawuala cave (10.0328 S, 120.1688 E) in (MGB-19656) and various small skull fragments, three vertebra a rich surficial bone deposit in a horizontal side-passage branching fragments and several costal fragments (figure 2; electronic back from the main downward-sloping passage approximately 10 m below the entrance. The side-passage terminates below the supplementary material, text S1 and figure S2). entrance chamber and contains a small talus cone including The Lewapaku juvenile mandible, representing dental further fossil material that has probably fallen through crevices wear stage dP3-dP4-A’, is of very small size even given its onto- from the entrance. Further fossils were found in a 50 cm test-pit genetic stage (height of horizontal ramus measured at level of in the horizontal entrance chamber of nearby Liang Minabuti anterior border of dental alveolus ¼ 43.8 mm; height of hori- cave (10.0298 S, 120.1718 E). Vertebrate material was collected by zontal ramus measured at level of anterior onset of ascending dry sieving. Exploration of 19 other caves across Sumba (electronic ramus ¼ 34.3 mm; total width of mandible at level of onset of supplementary material, table S1) only found unfossilized bones ascending ramii ¼ 108.3 mm (measurement M15 in [48]); of native bats and invasive murids that reached the eastern maximum transverse diameter of dextral horizontal ramus ¼ Lesser Sundas in the Late Holocene [20,24]. 40.4 mm (measurement M18); minimum transverse diameter Nearly all the Mahaniwa fossils represent an extensive sample of dextral horizontal ramus ¼ 15.2 mm (measurement M20)). of rodent bones. Abundant rodent mandibles and postcranial elements and a few cranial fragments (more than 3000 elements) As with size comparisons for the small adult mandible from were found in Liang Lawuala along with bat, cave swiftlet and Watumbaka, the sizes of the milk molars of the juvenile mand- reptile bones. We interpret this assemblage as probably deposited ible (right dp3 length ¼ 29.7 mm; width ¼ 19.9 mm; left dp4 by both natural accumulation (bone rain) from the entrance length ¼ 51.5 mm; width ¼ 25.8 mm) fall just within or chamber and concentration of small vertebrates by a taphonomic slightly below the size range of homologue elements of Downloaded from http://rspb.royalsocietypublishing.org/ on September 18, 2017

4 (a)(b)(c) rspb.royalsocietypublishing.org rc .Sc B Soc. R. Proc. 284

(d) (e) (g) 20171278 :

(h) ( f )

Figure 2. Pleistocene fossils from Lewapaku. (a–c) MGB-19650, Stegodon cf. sumbaensis.(a,c) Mandible, dorsal and lateral views; (b) isolated left and right m1 associated with mandible. (d) MGB-19665, Varanus cf. komodoensis tooth; (e) MGB-19670, rodent tibial diaphysis fragment; ( f ) MGB-19667, rodent lower incisor; (g) MGB-19669, small rodent proximal femur fragment; (h) MGB-19668, large rodent proximal femur fragment. (a,b) Scale bar, 2 cm; (c–g) scale bar, 1 cm.

S. sondaari (dp3 length ¼ 28–37.6 mm; width ¼ 15.7–19.6 mm; Similar-sized lower incisors are seen in Papagomys on Flores dp4 length ¼ 52.4–61.5 mm; width ¼ 22.3–29 mm; [48] and [23]. One proximal femur fragment (MGB-19668) is very G.D.v.d.B., unpublished data, 2017). The juvenile Lewapaku large, with a well-developed deltoid crest characteristic of mandible corresponds in expected size based on extrapolation murids. The other proximal femur fragment (MGB-19669) is from growth curves to the S. sumbaensis holotype mandible smaller and more incomplete. The tibial diaphysis fragment (electronic supplementary material, figure S3), and we interpret (MGB-19670), showing the divergence of fused tibia and it as probably conspecific or at least a member of the same line- fibula seen in most non-sciuromorph rodents [51], is also age. The Lewapaku mandible is also smaller than other juvenile referable to a giant murid. Wallacean Stegodon mandibles of similar wear stage (e.g. [20]). The Lewapaku tusk fragment is broken at both sides and measures 180 mm long. A cone-shaped cavity of the pulpa is (c) Varanus komodoensis Ouwens 1912 not present at either end, indicating that the fragment rep- The ziphodont varanid tooth from Lewapaku (MGB-19665; resents an intermediate tusk portion. It is slightly compressed figure 2) can be assigned to V. komodoensis. Its crown is strongly in cross-section, with a maximum diameter of 69.5 mm and compressed transversely and recurved backwards, with a sharp, minimum diameter of 60.1 mm; compression is probably delicately serrate posterior cutting edge, a height of 13.2 mm and partly due to post-mortem pressure, as various longitudinal anteroposterior diameter at crown base of 6.7 mm. Tooth size is cracks are present and the broken ends exhibit a crumbly, within the range of modern V. komodoensis (anteroposterior broken mass of dentine, so that the original diameter is diameter, 2.1–7.5 mm; crown height, 5.1–16.1 mm; n ¼ 87 therefore probably close to 65 mm. The tusk fragment is com- teeth from six individuals) [52]. parable in size to the largest tusk fragment known of S. sondaari (63 mm) [50]. Four tusk fragments from the slightly (d) Holocene Muridae larger Stegodon timorensis have a diameter of 81–85 mm Two large-bodied murids are present in the Mahaniwa sample. (G.D.v.d.B., unpublished data, 2017). These are markedly larger than introduced R. argentiventer and R. rattus from Sumba [25,43], and have distinctive mor- (b) Pleistocene Muridae phologies that differ from other Quaternary–Recent murids The Pleistocene rodent material from Lewapaku (figure 2) previously described from southeast Asia and Australasia (elec- is very fragmentary. The lower incisor (MGB-19667) is tronic supplementary material, text S2). Compared with other 31 mm long from tip to broken base and 2.57 mm wide. eastern Lesser Sunda murids, neither taxon exhibits the Downloaded from http://rspb.royalsocietypublishing.org/ on September 18, 2017

highly cuspidate complex molars or deep robust dentary of description of the Timor rodent, and this taxon differs in 5 Spelaeomys (Flores) or Coryphomys (Timor), and both have a simi- having molar laminar tubes that enter the bony structure of rspb.royalsocietypublishing.org lar higher number of molar roots (M1: 5, M2: 4, M3: 3, m1: 4, m2: the jaw, whereas in the new rodent the molar crown is still vis- 3, m3: 3) than these genera [12,13]. Instead, they show closer ible near the base of the lower molars in lateral profile. The molar and dentary morphology to Flores murids (Hooijeromys, mandible of the new rodent differs from that of Milimonggamys Komodomys, Papagomys, Paulamys), which are probably more in having a very high ascending ramus defined by a large closely related to each other than to other taxa [12]. coronoid process, and a wider and longer angular process; One of the new murids differs from these genera in having this morphology is much more similar to that seen in endemic a posterior cingulum on M3 (absent in Hooijeromys, Komodomys, Flores murids as well as ‘large murid genus B’ from Timor Papagomys and Paulamys); cusp t3 present on M2 and M3 (typi- and other large-bodied southeast Asian murids (e.g. Lenomys, cally absent in Papagomys and Paulamys) and reduced on M1 Mallomys) [12,34]. We describe this distinctive giant murid as

(relatively large in Hooijeromys and Komodomys, absent in the new genus and species Raksasamys tikusbesar (figure 3; B Soc. R. Proc. Paulamys); chevronate upper molar laminae (transverse in electronic supplementary material, text S2 and figure S5). Hooijeromys and Paulamys); M2 longer than wide (unlike Phylogenetic analysis was conducted to investigate the Hooijeromys); posterior margin of incisive foramina opposite affinities of Milimonggamys and Raksasamys within the biogeo- anterior M1 (more anterior in Hooijeromys, Papagomys and graphically complex wider murid radiations of southeast 284 Paulamys); anterolabial cuspid not disrupting transverse Asia and Australasia, using a comparative taxonomic sample anterior margin of m2 (versus Komodomys and Papagomys) of six other endemic murids from the Lesser Sunda Islands 20171278 : and posterior cingulid absent on m3 (versus Komodomys) (Hooijeromys, Komodomys, Papagomys, Paulamys, R. hainaldi, [12,23–26]. Its mandible has reduced coronoid and angular R. timorensis) and 17 other genera of Rattini, Hydromyini and processes that do not extend dorsally or posteriorly beyond Phloeomyini from the Sunda Shelf, the Philippines, Wallacea the level of the articular condyle; this mandible morphology and Sahul (Anisomys, Bunomys, Halmaheramys, Hydromys, is only similar to Komodomys among the endemic Flores Hyomys, Lenothrix, Leptomys, Macruromys, Mallomys, Maxomys, murid genera, but is otherwise similar to some southeast Parahydromys, Paruromys, Phloeomys, Pogonomys, Sundamys, Asian Rattini (e.g. Maxomys) [53]. It is smaller than Coryphomys, Taeromys, Uromys), and employing a matrix of 5966 characters Hooijeromys, Papagomys and Spelaeomys and larger than (25 cranial, mandibular and dental morphological characters, Komodomys and Paulamys; its body mass is estimated as and molecular data for four gene matrices taken from Fabre 251.2–336.3 g (electronic supplementary material, table S2). et al. [14]) (electronic supplementary material, text S3 and We describe it as the new genus and species Milimonggamys table S3). The phylogenetic placement of both Milimonggamys juliae (figure 3; electronic supplementary material, text S2 and Raksasamys in the majority consensus trees generated in and figure S4). these analyses was poorly resolved; only morphological data The other murid is much larger than Milimonggamys;itis are available for these taxa, and poor resolution is expected smaller than Coryphomys and Spelaeomys, larger than Hooijer- in investigations of murid relationships using morphological omys and close in size to P. theodorverhoeveni, with a body data due to homoplasy [13]. However, the majority consen- mass estimate of 468.2–680.5 g (electronic supplementary sus molecular þ morphology tree placed Milimonggamys, material, table S2). This giant murid has unusual, very Raksasamys, and all six other Lesser Sunda murids within high-crowned and strongly fluted molars with simplified the Rattini, and the separate parsimony and Bayesian analyses crown morphology and thin flattened upper molar laminae of the morphology-only dataset both returned a sister- that distinguish it from other eastern Lesser Sunda murids. group relationship for Raksasamys þ Papagomys (electronic It also differs in having transverse upper laminae (chevronate supplementary material, figure S6). in Komodomys, Milimonggamys and Papagomys); lacking a posterior cingulum on upper molars (versus Milimonggamys); (e) Varanus hooijeri Brongersma 1958 having M1 with central cusps t2, t5 and t8 similar in pro- Two varanid maxillaries (LL 2014/21–LL 2014/22; figure 3), portion to lingual cusps t1 and t4, but labial cusps t3, t6 each with four teeth, were also recovered at Liang Lawuala. and t9 extremely reduced compared with Komodomys and Teeth on both elements show the cylindrical shape and Papagomys; cusp t3 reduced on M1 and absent on M2 and blunt rounded bunodont crowns characteristic of posterior M3 (present in Hooijeromys, Komodomys and Milimonggamys); teeth of the extinct V. hooijeri from Flores [18] and cusp t7 present on M3 (absent in Hooijeromys, Komodomys, are otherwise only comparable to the extant frugivorous Milimonggamys, Papagomys and Paulamys); posterior margin Varanus bitatawa and Varanus olivaceus from Luzon, Philip- of incisive foramina markedly anterior to M1 (more posterior pines [54,55]. Tooth measurements are slightly smaller than in Komodomys and Milimonggamys); postpalatal margin oppo- for two V. hooijeri specimens from Flores; however, teeth in site middle of m3 (more posterior in Hooijeromys, Komodomys both Sumba maxillaries and both described Flores maxillaries and Paulamys); m2 approximately as wide as long (versus also vary in size, probably reflecting ontogenetic or intraspe- Milimonggamys and Paulamys); anterolabial cuspid disrupts cific variation [18] (electronic supplementary material, table transverse anterior margin of m2 (versus Hooijeromys, S4). We assign both elements to Varanus cf. hooijeri, revealing Milimonggamys and Paulamys); lacks well-expressed anterolabial a probable range extension for this extinct species, formerly cuspid on m3 (versus Hooijeromys, Komodomys, Milimonggamys, considered endemic to Flores. Papagomys and Paulamys); and almost always lacks anterolabial cusplet, posterolabial cusplet and posterior cingulid on lower molars (versus Komodomys, Milimonggamys, Papagomys and (f) Age estimates of the fossils Paulamys) [12,23–26]. It may be morphologically similar to the The holotype of S. sumbaensis originates from a fluvial layer at undescribed ‘large murid genus B’ from the Quaternary of Watumbaka that postdates the deposition, emergence and Timor [34]; however, proper comparison must await formal erosion of the lowest marine Terrace Complex I at Cape Downloaded from http://rspb.royalsocietypublishing.org/ on September 18, 2017

6 (a) (b) (c) (d) (e) rspb.royalsocietypublishing.org

( f ) rc .Sc B Soc. R. Proc. 284 (g) 20171278 :

(h)

(i) (j)

(k) (m)

(l) (n)

Figure 3. Holocene fossils from Mahaniwa. (a,c,k–l) Milimonggamys juliae gen. et sp. nov. (a) LL 2014/2, left maxillary toothrow; (b) LL 2014/1 (holotype), right mandibular toothrow; (k–l) LL 2014/7, right hemimandible; labial and lingual views. (b,d,m–n), Raksasamys tikusbesar gen. et sp. nov. (b) LL 2014/13, right maxillary toothrow; (d) LL 2014/17, right mandibular toothrow; (m–n) LL 2014/9 (holotype), right hemimandible; labial and lingual views. (e–j) Varanus cf. hooijeri.(e–g) LL 2014/21, right maxillary fragment; labial, lingual and occlusal views. (h–j) LL 2014/22, left maxillary fragment; occlusal, lingual and labial views. (a–d) Scale bar, 2 mm; (e–j) scale bar, 5 mm; (k–n) scale bar, 5 mm.

Laundi (because the fluvial deposits from which the Stegodon 3–20 m a.s.l.); U–Th dates indicate that this terrace complex mandible was recovered have cut down into Terrace Complex was formed during OI Stage 5, with coral ages corresponding I, they must be younger in age than the latter). Based on to the three main high sea-levels at 86 kyr BP, 105 kyr BP and elevation of approximately 20 m a.s.l., this sequence can 125 kyr BP [36,37]. As the terrace complex is polycyclic in be correlated with the lowest of the dated coral-reef nature, the lack of information about the exact stratigraphic terrace complexes near Cape Laundi (Terrace Complex I, at origin of the S. sumbaensis holotype within the Watumbaka Downloaded from http://rspb.royalsocietypublishing.org/ on September 18, 2017

outcrop precludes precise age estimation for this fossil, most southeast Asian islands continues to hinder rigorous 7 other than that it probably has a Late Pleistocene age younger comparison of inter-island faunal evolution and dispersal rspb.royalsocietypublishing.org than 125 kyr. across this biogeographically complex region, and it is possible Fossils at Lewapaku originate from a clayey layer that rests that taxa currently represented in modern or Quaternary unconformably on top of the deep marine sequence of the faunas from single Wallacean islands may have had much Waikabubak Formation. This marine sequence is also uncon- wider former distributions which have not yet been adequately formably overlaid by a series of Quaternary coral-reef sampled (cf. [58]). Future work on Quaternary microvertebrate terraces along the northern edge of the island at Cape faunas from Sumba may also provide important novel insights Laundi, which reach elevations of up to 475 m a.s.l.; these ter- into regional biogeography. However, our new baseline of past races have been extensively studied and dated by U–Th dating, diversity on Sumba allows new assessment of the island’s and document the youngest uplift phase of Sumba [36,37]. biogeographic affinities.

Uranium-series dating on a Stegodon molar fragment from V. komodoensis now has a relict distribution on Flores and B Soc. R. Proc. the Lewapaku assemblage (MGB-19652) did not produce nearby islands (Komodo, Rinca, Gili Motang, Padar), but had reliable results; the sample has experienced extensive uranium a wide Pliocene–Pleistocene distribution across both sides of loss, possibly because the fossil layer at Lewapaku is an aquifer Wallace’s Line, from Java to Australia [52]. This cosmopolitan (electronic supplementary material, text S1, figures S7, S8 and distribution limits its ability to provide a biogeographic 284 tables S5, S6). However, approximate age estimation of the signal. The species is currently unknown from the Quaternary Lewapaku site is possible by comparison of the elevation of record of Timor, where a distinctive unnamed larger extinct 20171278 : Quaternary terraces. At an elevation of 500–520 m a.s.l., the varanid has instead been recorded [52]; however, this apparent Lewapaku Basin is approximately 40 m higher than the highest absence may merely reflect limited sampling in this under- coral-reef terrace, which has been correlated with Oxygen Iso- studied region. Conversely, V. hooijeri is otherwise known tope Stage 27 (approx. 0.99 Ma) [56]. The fossil-bearing layer only from the Quaternary record of Flores [18,20,22], giving a that rests on the Waikabubak Formation must therefore be potential signal of biogeographic affinity between Flores and younger than approximately 1 Ma. However, it is difficult to Sumba. establish whether the fossil-bearing terrestrial layer began Milimonggamys and Raksasamys both share some dental and to accumulate directly following emergence of the basin, or mandibular characteristics with endemic murid genera from at a later stage. Flores and Timor, but exhibit other similarities to some other The Liang Lawuala and Liang Minabuti deposits contain southeast Asian murids (electronic supplementary material, characteristic Neolithic markers (pottery fragments, pig text S2 and table S3). Our phylogenetic analyses of the affinities bones), indicating a likely mid–late Holocene age. A small of Milimonggamys and Raksasamys generated trees that were amount (less than 1%) of murid postcrania at Liang Lawuala poorly resolved beyond being able to place these taxa and show signs of burning, suggesting possible consumption by the other sampled Lesser Sunda murids within the Rattini. humans (although see [33] for alternative potential explanations This poor resolution is probably due to the limited morpho- for burning on rodent bones). Bone samples from Liang Lawuala logical character data that were available for both new taxa, were submitted for accelerator mass spectrometer (AMS) 14C and to recognized problems of widespread homoplasy in dating at the 14CHRONO Centre, Queen’s University, Belfast murid craniodental characters [13]. Fuller understanding of (UK), and calibrated with OxCal 4.2 [57]. Three AMS 14Cdates regional Quaternary biogeography and evolution must also were obtained for Liang Lawuala: a date of 2167 + 24 BP await description and phylogenetic analysis of Timor’s extinct (¼279–119 calBC) from charcoal from 5–10 cm depth (UBA- rodent fauna [13,34], and future attempts to extract ancient 30134), a direct date of 1889 + 33 BP (¼54–222 calAD) from a DNA from Quaternary rodent material from the Lesser Milimonggamys mandible from 0–5 cm depth (UBA-30135) Sundas, which may not be feasible because of their preser- and a direct date of 3507 + 38 BP (¼1935–1700 calBC) from a vation under tropical environmental conditions. Rigorous large rodent scapula surface find probably referable to Raksasa- assessment of phylogenetic and biogeographic affinities of mys (UBA-30138). Two further rodent specimens (Raksasamys Sumba’s Quaternary rodent fauna, and whether its endemic femur, mandible; UBA-30135, 30136) did not return AMS rodents represent a single within-island evolutionary radiation dates because of insufficient collagen yield. or multiple overwater colonizations from one or more islands, is thus not yet possible. However, both of our morphology- only phylogenetic analyses returned a sister-group relationship for Raksasamys þ Papagomys, providing some support for a 4. Discussion close biogeographic relationship between the endemic Our fossil finds provide the first detailed understanding of murids of Sumba and Flores, although in the absence of com- Quaternary vertebrate diversity and turnover on Sumba, a parative character data for any endemic Timor rodents other large Indonesian island for which the terrestrial fossil record than R. timorensis. was almost totally unknown, and which is situated in a bio- Sumba is situated less than 50 km south of Flores, and geographically complex and important region. We reveal that extensive inter-ocean transport from the Pacific Ocean to the Sumba’s Quaternary vertebrate fauna was similar in diversity Indian Ocean via the Indonesian Throughflow, a major cross- and composition to the unusual ‘unbalanced’ fauna on roads of global ocean circulation, occurs southwards past Flores, containing not only a pygmy Stegodon but also Flores through the Sape and Sumba straits immediately north Komodo dragons, another unusual and apparently regionally of Sumba [59]. Overwater ‘waif dispersal’ from Flores to endemic varanid, and large-bodied murids. It is therefore likely Sumba is therefore a biologically plausible chance event, pro- that similar assemblages probably characterized Quaternary viding a likely explanation for the apparent biogeographic terrestrial ecosystems on many or all of the larger Lesser affinity seen between some components of the described Sunda Islands. The poor available Quaternary record for Quaternary faunas from these islands. However, observed Downloaded from http://rspb.royalsocietypublishing.org/ on September 18, 2017

differences in size and diversity in different taxa between Mahaniwa and elsewhere on Sumba reported that ‘giant rats’ 8 Sumba, Flores and other Lesser Sunda Islands—notably larger than R. rattus still lived in the island’s forests, and rspb.royalsocietypublishing.org lower rodent species richness and maximum body size on large rodents feature in local myths recounted to us during Sumba versus Flores and Timor, smaller size of Stegodon fieldwork in 2014. Southeast Asia contains the world’s highest sumbaensis compared with most other Stegodon species, and levels of globally threatened mammals [63], and further field- possibly also inter-island differences in varanid faunas—may work to assess the possible survival of Milimonggamys or be explicable on biogeographic grounds and parallel evolution Raksasamys could represent an important conservation priority. rather than close regional evolutionary history, as these pat- Our research constitutes an important new step towards terns are consistent with body size trends and species clarifying the evolutionary history and Quaternary diversity richness increasing with land area in other island systems [3,5]. of the Wallacean vertebrate fauna, and future palaeontological Discovery of Pleistocene and Holocene faunas on Sumba research on Sumba will no doubt provide further insights into that may span much or most of the last 1 Myr, including taxa vertebrate evolution in the Lesser Sundas. Whereas most B Soc. R. Proc. directly dated to the Late Holocene, provides a temporal frame- other components of Flores’ Quaternary vertebrate fauna were work to reconstruct Quaternary vertebrate turnover compared found at Liang Bua cave in the mid-twentieth century, exca- with faunal sequences for Flores. If the Lewapaku deposit vations took place at this site for almost 50 years before is close to 1 million years old based on correlation with discovery of H. floresiensis [4]. We therefore consider it likely 284 coral-reef terraces at Cape Laundi, this provides a similar first that new vertebrate taxa remain to be discovered in Sumba’s occurrence for pygmy Stegodon and V. komodoensis on Sumba Quaternary fossil record. Most intriguingly, given other simi- 20171278 : to the oldest record of these taxa on Flores; rodents first larities between the distinctive Quaternary faunas from appear on Flores at approximately 800 ka, although this is Sumba and Flores, there is no obvious biogeographic reason probably a taphonomic artefact rather than true absence from to suppose that regionally endemic hominin species would older horizons [16,22,60,61]. However, the lack of dated Pleis- have been present on Flores but not Sumba. Indeed, the possi- tocene deposits on Sumba prevents assessment of whether bility of ancient hominin occurrence on Sumba has already subsequent faunal turnover occurred at the Early to Middle been raised by the apparent presence of Palaeolithic artefacts Pleistocene transition as seen on Flores [22,60,61]. St. sumbaen- in Pleistocene sediments [64]. We strongly encourage further sis and V. komodoensis would not be expected to occur in an owl investigation of Sumba’s Quaternary preservational environ- roost deposit, so that their absence from the Mahaniwa fossil ments, which may well provide exciting new lessons for sample cannot help to constrain the likely extinction timing understanding both island biogeographyand human evolution. for either species. The limited, non-diagnostic rodent material from Lewapaku also prevents assessment of whether Data accessibility. Sumba’s Pleistocene murids were part of the same endemic The data supporting this article are described in full in the electronic supplementary material. lineage as Milimonggamys and Raksasamys; phylogenetic conti- Authors’ contributions. S.T.T. and G.D.v.d.B. designed research; S.T.T., nuity is likely, but local extinction and recolonization by G.D.v.d.B., J.J.C., J.H., T.E.J., N.C., I.K., E.S. and U.P. collected data; different rodent lineages during the Late Quaternary is also S.T.T., G.D.v.d.B., J.J.C., T.G. and A.D. analysed data; and S.T.T. possible, a phenomenon also shown regionally by stratigraphic and G.D.v.d.B. wrote the paper. All authors gave final approval for turnover of different colonizing Stegodon lineages on Flores publication. [20,22]. Competing interests. We declare we have no competing interests. Sumba and Flores also both retained large murids and Funding. Funding was provided by a Royal Society University Varanus cf. hooijeri into the late Holocene [20,22,23]. The current Research Fellowship (UF080320/130573) and an Australian Research Council Future Fellowship (FT100100384). status of Sumba’s endemic murids is unknown, as little Acknowledgements. For permission and support for fieldwork, we thank the mammal survey work has been conducted on the island. Indonesian State Ministry of Research and Technology (RISTEK), P. theodorverhoeveni and Sp. florensis died out in the Holocene S. Permanandewi (former Head of Geological Survey Institute, Ban- on Flores, and Milimonggamys and Raksasamys may now also dung), Y. Kusumahbrata (former Head of Geology Museum Bandung), be extinct because of habitat loss, hunting and/or competition Inen Rochaeni, Mr Hidayat and Mrs Yuhartini (Geological Survey Insti- with introduced species in recent millennia. However, one tute), Tatang (Geology Museum Bandung), Jublina Tode Solo (Culture and Tourism Department NTT Province, Kupang), the staff of the Interior giant murid, P. armandvillei, survives on Flores. Most of Ministry Office (Waingapu), Laura D’Arcy, Andjar Rafiastanto, Astri Sumba’s forests have been burnt or cleared for agriculture or Dewi, Dudy Nugroho and Hannah Timmins (ZSL Indonesia office, firewood extraction and replaced by grassland [62], but Bogor), the family of the late Kornelis Ngongo Mbani, Kornelis Ngo Sumba has a much lower human population density than Tanggu, Andreas Landi, Marten Lende and the inhabitants of Mahaniwa Flores and has experienced almost no human population village. Access to museum collections was facilitated by Tony Djubian- tono (National Centre for Archaeology, Jakarta), and Louise Tomsett growth since the mid-nineteenth century [8], suggesting that and Roberto Portela Miguez (Natural History Museum, London). Further anthropogenic pressures may not have been so severe as to logistical and technical support was provided by Gregory Forth, Retang drive extinction of the native murids. Indeed, local people in Wohangara, Thomas Brown and Pierre-Henri Fabre.

References

1. Wallace AR. 1860 On the zoological geography of Nature 513, 543–546. (doi:10.1038/ USA 98, 14 518–14 523. (doi:10.1073/pnas. the Malay Archipelago. J. Linn. Soc. 4, 172–184. nature13739) 251548698) (doi:10.1111/j.1096-3642.1860.tb00090.x) 3. Burness GP, Diamond J, Flannery T. 2001 4. van der Geer A, Lyras G, de Vos J, 2. Helmus MR, Mahler DL, Losos JB. 2014 Dinosaurs, dragons, and dwarfs: the evolution Dermitzakis M. 2010 Evolution of island Island biogeography of the Anthropocene. of maximal body size. Proc. Natl Acad. Sci. mammals: adaptation and extinction of placental Downloaded from http://rspb.royalsocietypublishing.org/ on September 18, 2017

mammals on islands. Chichester, UK: Wiley- Flores, Indonesia. Quat. Int. 182, 16–48. (doi:10. interactions and owl roosting at Tron Bon Lei, Alor 9 Blackwell. 1016/j.quaint.2007.02.001) Island, eastern Indonesia. J. Island Coastal Archaeol, rspb.royalsocietypublishing.org 5. Van der Geer AAE, van den Bergh GD, Lyras GA, 20. van den Bergh GD et al. 2009 The Liang Bua faunal 1–17. (doi:10.1080/15564894.2017.1285834) Prasetyo UW, Awe Due R, Setiyabudi E, Drinia H. remains: a 95 k.yr. sequence from Flores, east 34. Glover I. 1986 Archaeology in eastern Timor, 2016 The effect of area and isolation on insular Indonesia. J. Hum. Evol. 57, 527–537. (doi:10. 1966–67. Terra Australis 11, 1–241. dwarf proboscideans. J. Biogeogr. 43, 1656–1666. 1016/j.jhevol.2008.08.015) 35. Dennell W, Louys J, O’Regan HJ, Wilkinson DM. (doi:10.1111/jbi.12743) 21. Meijer HJM, Awe Due R. 2010 A new species of 2014 The origins and persistence of Homo 6. Steadman DW. 2006 Extinction and biogeography of giant marabou stork (Aves: Ciconiiformes) from the floresiensis on Flores: biogeographical and ecological tropical Pacific birds. Chicago, IL: University of Pleistocene of Liang Bua, Flores (Indonesia). perspectives. Quat. Sci. Rev. 96, 98–107. (doi:10. Chicago Press. Zool. J. Linn. Soc. 160, 707–724. (doi:10.1111/j. 1016/j.quascirev.2013.06.031) 7. Faurby S, Svenning JC. 2015 Historic and prehistoric 1096-3642.2010.00616.x) 36. Pirazzoli PA, Radtke U, Hantoro WS, Jouannic C,

human-driven extinctions have reshaped global 22. Meijer HJM, van den Hoek Ostende LW, van den Hoang CT, Causse C, Borel-Best M. 1991 Quaternary B Soc. R. Proc. mammal biodiversity patterns. Divers. Distrib. 25, Bergh GD, de Vos J. 2010 The fellowship of the raised coral-reef terraces on Sumba Island, 1155–1166. (doi:10.1111/ddi.12369) hobbit: the fauna surrounding Homo floresiensis. Indonesia. Science 252, 1834–1836. (doi:10.1126/ 8. Monk KA, de Fretes Y, Reksodiharjo-Lilley G. 1997 J. Biogeogr. 37, 995–1006. (doi:10.1111/j.1365- science.252.5014.1834) The ecology of Nusa Tenggara and Maluku. Jakarta, 2699.2010.02308.x) 37. Pirazzoli PA, Radkte U, Hantoro WS, Jouannic C, Indonesia: Periplus Editions. 23. Locatelli E, Awe Due R, van den Bergh GD, van den Hoang CT, Causse C, Borel Best M. 1993 A one 284 9. Michaux B. 2010 Biogeology of Wallacea: Hoek Ostende LW. 2012 Pleistocene survivors and million-year-long sequence of marine terraces on 20171278 : geotectonic models, areas of endemism, and natural Holocene extinctions: the giant rats from Liang Bua Sumba Island, Indonesia. Mar. Geol. 109, 221–236. biogeographical units. Biol. J. Linn. Soc. 101, (Flores, Indonesia). Quat. Int. 281, 47–57. (doi:10. (doi:10.1016/0025-3227(93)90062-Z) 193–212. (doi:10.1111/j.1095-8312.2010.01473.x) 1016/j.quaint.2012.04.005) 38. Wensink H, van Bergen MJ. 1995 The tectonic 10. van den Bergh GD, de Vos J, Sondaar PY. 2001 The 24. Locatelli E, Awe Due R, van den Hoek Ostende LW. placement of Sumba in the Sunda-Banda Arc: Late Quaternary palaeogeography of mammal 2015 Middle-sized murids from Liang Bua (Flores, paleomagnetic and geochemical evidence from the evolution in the Indonesian archipelago. Indonesia): insular endemics, human introductions early Miocene Jawila volcanics. Tectonophys. 250, Palaeogeogr. Palaeoclimatol. Palaeoecol. 171, and palaeoenvironment. Palaeobio. Palaeoenv. 95, 15–30. (doi:10.1016/0040-1951(95)00048-5) 385–408. (doi:10.1016/S0031-0182(01)00255-3) 497–512. (doi:10.1007/s12549-015-0204-1) 39. Rutherford E, Burke K, Lytwyn J. 2001 Tectonic 11. Louys J, Price GJ, O’Connor S. 2016 Direct dating of 25. Musser GG. 1980 A new genus of murid rodent history of Sumba Island, Indonesia, since the Late Pleistocene stegodon from Timor Island, East Nusa from the Komodo Islands in Nusatenggara, Cretaceous and its rapid escape into the forearc in Tenggara. PeerJ 4, e1788. (doi:10.7717/peerj.1788) Indonesia. J. Mammal. 61, 395–413. (doi:10.2307/ the Miocene. J. Asian Earth Sci. 19, 453–479. 12. Musser GG. 1981 The giant rat of Flores and its 1379834) (doi:10.1016/S1367-9120(00)00032-8) relatives east of Borneo and Bali. Bull. Am. Mus. 26. Kitchener DJ, How RA. 1991 Paulamys sp. cf. P. naso 40. Coates BJ, Bishop KD. 1997 A guide to the birds of Nat. Hist. 169, 67–176. (Musser, 1981) (Rodentia: Muridae) from Flores Wallacea. Alderley, Australia: Dove Publications. 13. Aplin KP, Helgen KM. 2010 Quaternary murid Island, Nusa Tenggara, Indonesia – description 41. Olsen J, Trost S, Myers SD. 2009 Owls on the rodents of Timor. Part I: new material of from a modern specimen and a consideration of its island of Sumba, Indonesia. Austr. Field Ornithol. Coryphomys buelheri Schaub, 1937, and description phylogenetic affinities. Rec. West Aust. Mus. 15, 26, 2–14. of a second species of the genus. Bull. Am. Mus. 171–189. 42. Dammerman KW. 1928 On the mammals of Sumba. Nat. Hist. 341, 1–80. (doi:10.1206/692.1) 27. Kitchener DJ, How RA, Maharadatunkamsi. 1991 A Treubia 10, 299–315. 14. Fabre PH et al. 2013 A new genus of rodent from new species of Rattus from the mountains of west 43. Musser GG. 1972 Identities of taxa associated with Wallacea (Rodentia: Muridae: Murinae: Rattini), and Flores, Indonesia. Rec. West Aust. Mus. 15, 611–626. Rattus rattus (Rodentia, Muridae) of Sumba Island, its implication for biogeography and Indo-Pacific 28. Zijlstra JE, van den Hoek Ostende LW, Awe Due R. Indonesia. J. Mammal. 53, 861–865. (doi:10.2307/ Rattini systematics. Zool. J. Linn. Soc. 169, 408– 2008 Verhoeven’s giant rat of Flores (Papagomys 1379220) 447. (doi:10.1111/zoj.12061) theodorverhoeveni, Muridae) extinct after all? 44. Sartono S. 1979 The discovery of a pygmy Stegodon 15. Brown P, Sutikna T, Morwood MJ, Soejono RP, Contrib. Zool. 77, 25–31. from Sumba, East Indonesia: an announcement. Jatmiko WSE, Awe Due R. 2004 A new small-bodied 29. Hooijer DA. 1972 Stegodon trigonocephalus florensis Mod. Quat. Res. SE Asia 5, 57–63. hominin from the Late Pleistocene of Flores, Hooijer and Stegodon timorensis Sartono from the 45. Setiyabudi E, Kurniawan I, van den Bergh GD. 2012 Indonesia. Nature 431, 1055–1061. (doi:10.1038/ Pleistocene of Flores and Timor. I & II. Proc. Fossils of Stegodon and Varanus komodoensis nature02999) Koninklijke Ned. Akad. Wetensc. B 75, 12–33. Sumba and Flores: a Pleistocene landbridge? In IAGI 16. Brumm A, Aziz F, van den Bergh GD, Morwood MJ, 30. Hooijer DA. 1972 Varanus (Reptilia, Sauria) from the Digital Proc. of 2012 Annual Meeting (2012-SS-31). Moore MW, Kurniawan I, Hobbs DR, Fullagar R. Pleistocene of Timor. Zool. Med. 47, 445–448. Indonesia: Ikatan Ahli Geologi Indonesia. 2006 Early stone technology on Flores and its 31. Flannery TF, Bellwood P, White JP, Moore A, Boeadi 46. Effendi AC, Apandi T. 1993 Geological map of implications for Homo floresiensis. Nature 441, NG. 1995 Fossil marsupials (Macropodidae, Waikabubak and Waingapu, Nusatenggara. Scale 624–628. (doi:10.1038/nature04618) Peroryctidae) and other mammals of Holocene age 1:250,000. Bandung, Indonesia: Geological Research 17. Sutikna T et al. 2016 Revised stratigraphy and from Halmahera, north Moluccas, Indonesia. and Development Centre. chronology for Homo floresiensis at Liang Bua in Alcheringa 19, 17–25. (doi:10.1080/0311551 47. Andrews P. 1990 Owls, caves and fossils. Chicago, IL: Indonesia. Nature 532, 366–369. (doi:10.1038/ 9508619095) University of Chicago Press. nature17179) 32. Aplin K, Pasveer J. 2005 Mammals and other 48. van den Bergh GD. 1999 The Late Neogene 18. Brongersma LD. 1958 On an extinct species of the vertebrates from late Quaternary archaeological sites elephantoid-bearing faunas of Indonesia and their genus Varanus (Reptilia, Sauria) from the island of on Pulau Kobroor, Aru Islands, Eastern Indonesia. In palaeozoogeographic implications. Scr. Geolog. 117, Flores. Zool. Med. 36, 113–125. The archaeology of the Aru Islands, Eastern 1–419. 19. van den Bergh GD, Awe Due R, Morwood MJ, Indonesia (eds S O’Connor, M Spriggs, P Veth), 49. Beden M. 1979 Les e´lephants (Loxodonta et Sutikna T, Jatmiko P, Wahyu Saptomo E. 2008 The pp. 41–62. Canberra, Australia: Pandanus Books. Elephas) d’Afrique orientale: syste´matique, youngest Stegodon remains in southeast Asia from 33. Hawkins S, Samper Carro SC, Louys J, Aplin K, phyloge´nie, inte´reˆt biochronologique. Doctoral the Late Pleistocene archaeological site Liang Bua, O’Connor S. 2017 Human palaeoecological thesis, University of Poitiers, France. Downloaded from http://rspb.royalsocietypublishing.org/ on September 18, 2017

50. Aziz F, van den Bergh GD, Morwood MJ, Hobbs D, lizard reveals a hidden biogeographic boundary and circulation in the Indonesian seas: 1/128 global 10 Kurniawan I, Collins JG. 2009 Excavations at Tangi a novel flagship species for conservation. Biol. Lett. HYCOM and the INSTANT observations. Dynam. rspb.royalsocietypublishing.org Talo, central Flores, Indonesia. In Pleistocene 6, 654–658. (doi:10.1098/rsbl.2010.0119) Atmos. Ocean 50, 275–300. (doi:10.1016/j. geology, palaeontology and archaeology of the Soa 55. Welton LJ, Siler CD, Diesmos AC, Diesmos MLL, dynatmoce.2010.04.002) Basin, central Flores, Indonesia (eds F Aziz, MJ Lagat RD, Causaren RM, Brown RM. 2012 Genetic 60. Morwood MJ, O’Sullivan PB, Aziz F, Raza A. 1998 Morwood, GD van den Bergh), pp. 41–58. identity, geographic ranges, and major distribution Fission-track ages of stone tools and fossils on Bandung, Indonesia: Geological Survey Centre records for frugivorous monitor lizards of Luzon the east Indonesian island of Flores. Nature 392, Bandung, Special Publication 36. Island, Philippines. Herpetol. Rev. 43, 226–230. 173–176. (doi:10.1038/32401) 51. Moss ML. 1977 A functional analysis of fusion of the 56. Bard E, Jouannic C, Hamelin B, Pirazzoli PA, Arnold 61. Brumm A et al. 2016 Age and context of the oldest tibia and fibula in the rat and mouse. Acta Anatom. M, Faure G, Sumosusastro P. 1996 Pleistocene sea known hominin fossils from Flores. Nature 534, 97, 321–332. (doi:10.1159/000144749) levels and tectonic uplift based on dating of corals 249–253. (doi:10.1038/nature17663)

52. Hocknull SA, Piper PJ, van den Bergh GD, Awe Due from Sumba Island, Indonesia. Geophys. Res. Lett. 62. Wikramanayake E, Dinerstein E, Loucks CJ. 2000 B Soc. R. Proc. R, Morwood MJ, Kurniawan I. 2009 Dragon’s 23, 1473–1476. (doi:10.1029/96GL01279) Terrestrial ecoregions of the Indo-Pacific: a paradise lost: palaeobiogeography, evolution and 57. Bronk RC. 2013 OxCal 4.2 manual. See http://c14. conservation assessment. Washington, DC: Island extinction of the largest-ever terrestrial lizards arch.ox.ac.uk/oxcalhelp/hlp_contents.html. Press. (Varanidae). PLoS ONE 4, e7241. (doi:10.1371/ 58. Steadman DW. 2006 An extinct species of tooth- 63. Schipper J et al. 2008 The status of the world’s land journal.pone.0007241) billed pigeon (Didunculus) from the Kingdom of and marine mammals: diversity, threat, and 284 53. Musser GG. 1991 Sulawesi rodents: descriptions of Tonga, and the concept of endemism in insular knowledge. Science 322, 225–230. (doi:10.1126/ 20171278 : new species of Bunomys and Maxomys (Muridae, landbirds. J. Zool. 268, 233–241. (doi:10.1111/j. science.1165115) Murinae). Am. Mus. Novit. 3001, 1–41. 1469-7998.2005.00010.x) 64. Simanjuntak T, Se´mah F, Gaillard C. 2010 The 54. Welton LJ, Siler CD, Bennett D, Diesmos A, Roy 59. Metzger EJ, Hurlburt HE, Xu X, Shriver JF, palaeolithic in Indonesia: nature and chronology. Duya M, Dugay R, Rico ELB, Van Weerd M, Brown Gordon AL, Sprintall J, Susanto RD, van Quat. Int. 223–224, 418–421. (doi:10.1016/j. RM. 2010 A spectacular new Philippine monitor Aken HM. 2010 Simulated and observed quaint.2009.07.022)